Official Qwen blog: Qwen3.8-Max is most capable Qwen family model; first Max-class open-weight release planned next week. Built on Qwen3.5 architecture; 2.4T parameters with 95B active. Focus on coding, work, research, long-horizon tasks. Available via QwenCloud APIs; weights on Hugging Face and ModelScope next week. Compatible with OpenAI Chat Completions and Anthropic API protocols for coding agents.